Things that are still uncomfortable in my experience that should scare people considering Haskell for their business:
* relability of hackage / stackage infrastructure. Could they fall off the face of the Earth one day and your business is now hosed?
* is there enough momentum in the community to keep the ecosystem going if something happens to the big contributors? How high is the bus factor for the ecosystem?
* finding qualified people is non-trivial. It's doable to find good software engineers with crappy Haskell skills or bad software engineers with great Haskell skills, finding both is a challenge.
* building large Haskell projects still takes forever / a lot of memory, breaks a lot of CI out there requiring manual set up of Jenkins and friends, not a lot of documentation and success stories out there to use for reference. GHC seems to get slower with every iteration.
* small community, not clear if the support will be there if something goes south in your project. With Java and friends you have years of accumulated documentation on how to troubleshoot anything on the internet, an infinity of experts who can look into your issues. With Haskell, not so much.
Will see if I can think of more of these.